I don’t have work to present on workshop sessions. So I was hanging around today, choosing some of the talks to listen to.

Invited Talk: Learning a Deep Convolutional Network for Image Super-resolution, Chen Change Loy (CUHK, Hong Kong)

This is the most valuable talk in my opinion.

Invited Talk by Xiaogang Wang on Deep Learning about Face
Deep Learning Face Representation by Joint Identification-Verification

The accent of Xiaogang is quite Chinglish, much worse than Chen Change Loy. Even so, Xiaogang’s work are great. It seems that Xiaogang and Shuicheng Yan have a close relationship.

Invited Talk: Deep Learning for Image Classification - How Far Have We Come? Hanlin Goh (Astar, Singapore)

Invited Talk: Towards real time CNN trackers, Yi Li (NICTA, Autralia)

Interesting guy, and fantastic work.

The final oral session by two Chinese students(from Tongji and CAS) are awful. The poster session, however is better and interesting, in my view. It seems quite emergency for Chinese students to improve their skills on giving talks.

It is quite ironic that most of the invited speakers were talking mainly about their work on CVPR 2014 and ECCV 2014, while their work on this ACCV seems not as excellent.

Uptate on Nov. 3:
It turns out that some of the poster session papers in the W3 - Deep Learning on Visual Data will give oral presentation on the main conference track.
I’m talking about:

Deep Representations to Model User ‘Likes’, Sharath Chandra Guntuku, Nanyang Technological University, Singapore; Joey Tianyi Zhou, Nanyang Technological University, Singapore; Sujoy Roy, Institute for Infocomm Research, Singapore; Lin Weisi, Nanyang Technological University, Singapore; Ivor Tsang, University of Technology, Australia.